基于现代数字信号处理理论提高SAR图像分辨率
|
摘要
随着合成孔径雷达越来越广泛应用,人们对雷达图像的分辨率的要求也越来越高。但在通过改进和更新硬件技术来提高图像分辨率已经成为一个难以突破的瓶颈的今天,如何利用数字信号处理技术来提高图像的分辨率就显得十分重要了。
对于合成孔径雷达(SAR)系统来说,雷达与目标之间的相对运动,既是成像的前提条件,也是导致图像模糊的根源。一方面,运动提供了成像所需要的含有多普勒信息的回波信号:另一方面,运动的不确定性造成的相位误差,又是高分辨率的制约因素。随着对分辨率要求的提高,在孔径合成时间内,对天线相位中心与成像目标之间运动补偿精度的要求也相应地提高了。即便是经过高精度的运动补偿,由于成像算法上的近似,以及大气传播效应等产生的相位误差仍会造成散焦,导致图像的恶化。因此,为得到聚焦良好的图像,在利用惯导数据进行运动补偿之后,高分辨成像往往还要求从SAR回波数据中估计并补偿残余的相位误差,这一过程称为自聚焦。
在合成孔径雷达成像算法中,利用自聚焦算法估计雷达回波信号的多普勒参数来校正相位误差是必不可少的一个步骤,相位梯度自聚焦(PGA)算法因其较好的稳健性而得到广泛应用。经典的相位梯度自聚焦算法选择能量较高距离单元中的最强散射点作为特显点进行多普勒参数估计。但由于背景杂波和相邻目标的干扰,采用这种策略挑选出的强散射点往往不是孤立的,这将导致相位误差估计的准确性下降。
本文在从理论上分析了经典相位梯度自聚焦算法不足的基础上,提出了一种改进算法:将幅度较大的散射点与周围背景区的均值相比较,以该比值作为该像素点质量的评估,选取比值较大的点作为特显点,这样选取的特显点一般都是孤立的。利用这种孤立特显点来估计相位误差,其准确性会得到很大提高,图像的聚焦效果也得到了改善,尤其是可以将有强背景杂波情况下相位梯度自聚焦算法的聚焦效果大幅度地提高。
论文最后还通过仿真对改进算法进行了验证,并分析了实验结果与改进算法的性能。
Nowadays, the Synthetic aperture radar (SAR) is applied increasingly in the more and more fields, and so, the requirement of developing high-resolution of the SAR is more and more intense.But in today when it is a bottleneck to promote the resolution with improving and updating the hardware configuration, it seems very important how to improve the resolution by the technology of advanced digital signal processing.
For SAR, imaging is based on the relative movement of the radar and the target, but it is the relative movement that causes the imaging indistinction. On the one hand, the Doppler message in the echo signal is provided by the relative movement, and on the other hand, the imaging high-resolution is restricted by the phase error issued from the nondeterminacy of the relative movement. Along with the requirement of the high-resolution, the accuracy of compensating the motion errors must to be improved. Yet even this, the phase error brought by the atmospheric propagation error and the approximating of imaging algorithms is still a one of the dominating reasons decreasing imaging quality and causing image out-of-focus. So, in order to get a clear image, estimating and compensating the phase error from the echo signal of SAR after compensating the motion errors by high-precision INS which is called autofocus is essential.
In all imaging algorithms of SAR, using autofocus algorithms to estimate the Doppler parameters to compensate the phase errors is a crucial step. And in these autofocus algorithms, the phase gradient autofocus (PGA) algorithm is frequently used because of it's robustness. In PGA algorithms, the powerfull scatter in range unit possessing more energy is selected to estimate the Doppler parameters as the dominant scatter. But if the scatter selected in PGA algorithm is not isolated because of the interference caused by background noise and adjacent target, the accuracy of the phase error estimation will be degraded.
In this dissertation, after analyzing the limitation of the classical PGA algorithms, an improved PGA algorithm is presented to select an isolated scatter. And in this improved PGA algorithm, each rather too powerfull pixel in the image is evaluated, and then the high quality scatters are selected for the phase error estimation.
A simulational experimentation is designed to verify the performance of the improved algorithm in the end of this dissertation. And the results of experimentation show that the improved PGA algorithm can elevate the accuracy of the phase error estimation.
对于合成孔径雷达(SAR)系统来说,雷达与目标之间的相对运动,既是成像的前提条件,也是导致图像模糊的根源。一方面,运动提供了成像所需要的含有多普勒信息的回波信号:另一方面,运动的不确定性造成的相位误差,又是高分辨率的制约因素。随着对分辨率要求的提高,在孔径合成时间内,对天线相位中心与成像目标之间运动补偿精度的要求也相应地提高了。即便是经过高精度的运动补偿,由于成像算法上的近似,以及大气传播效应等产生的相位误差仍会造成散焦,导致图像的恶化。因此,为得到聚焦良好的图像,在利用惯导数据进行运动补偿之后,高分辨成像往往还要求从SAR回波数据中估计并补偿残余的相位误差,这一过程称为自聚焦。
在合成孔径雷达成像算法中,利用自聚焦算法估计雷达回波信号的多普勒参数来校正相位误差是必不可少的一个步骤,相位梯度自聚焦(PGA)算法因其较好的稳健性而得到广泛应用。经典的相位梯度自聚焦算法选择能量较高距离单元中的最强散射点作为特显点进行多普勒参数估计。但由于背景杂波和相邻目标的干扰,采用这种策略挑选出的强散射点往往不是孤立的,这将导致相位误差估计的准确性下降。
本文在从理论上分析了经典相位梯度自聚焦算法不足的基础上,提出了一种改进算法:将幅度较大的散射点与周围背景区的均值相比较,以该比值作为该像素点质量的评估,选取比值较大的点作为特显点,这样选取的特显点一般都是孤立的。利用这种孤立特显点来估计相位误差,其准确性会得到很大提高,图像的聚焦效果也得到了改善,尤其是可以将有强背景杂波情况下相位梯度自聚焦算法的聚焦效果大幅度地提高。
论文最后还通过仿真对改进算法进行了验证,并分析了实验结果与改进算法的性能。
Nowadays, the Synthetic aperture radar (SAR) is applied increasingly in the more and more fields, and so, the requirement of developing high-resolution of the SAR is more and more intense.But in today when it is a bottleneck to promote the resolution with improving and updating the hardware configuration, it seems very important how to improve the resolution by the technology of advanced digital signal processing.
For SAR, imaging is based on the relative movement of the radar and the target, but it is the relative movement that causes the imaging indistinction. On the one hand, the Doppler message in the echo signal is provided by the relative movement, and on the other hand, the imaging high-resolution is restricted by the phase error issued from the nondeterminacy of the relative movement. Along with the requirement of the high-resolution, the accuracy of compensating the motion errors must to be improved. Yet even this, the phase error brought by the atmospheric propagation error and the approximating of imaging algorithms is still a one of the dominating reasons decreasing imaging quality and causing image out-of-focus. So, in order to get a clear image, estimating and compensating the phase error from the echo signal of SAR after compensating the motion errors by high-precision INS which is called autofocus is essential.
In all imaging algorithms of SAR, using autofocus algorithms to estimate the Doppler parameters to compensate the phase errors is a crucial step. And in these autofocus algorithms, the phase gradient autofocus (PGA) algorithm is frequently used because of it's robustness. In PGA algorithms, the powerfull scatter in range unit possessing more energy is selected to estimate the Doppler parameters as the dominant scatter. But if the scatter selected in PGA algorithm is not isolated because of the interference caused by background noise and adjacent target, the accuracy of the phase error estimation will be degraded.
In this dissertation, after analyzing the limitation of the classical PGA algorithms, an improved PGA algorithm is presented to select an isolated scatter. And in this improved PGA algorithm, each rather too powerfull pixel in the image is evaluated, and then the high quality scatters are selected for the phase error estimation.
A simulational experimentation is designed to verify the performance of the improved algorithm in the end of this dissertation. And the results of experimentation show that the improved PGA algorithm can elevate the accuracy of the phase error estimation.
引文
[1]皮亦鸣、杨建宇、付毓生.合成孔径雷达成像原理.成都:电子科技大学出版社,2007.3
[2]王正明、朱炬波等著.SAR图像提高分辨率技术.北京:科学出版社,2006.10
[3]张直中.雷达信号的选择与处理.北京:国防工业出版社,1989
[4]J.Li,P.Stoica.An Adaptive Filtering Approach to Spectral Estimation and SAR Imaging.IEEE Signal Processing,June 1996,Vol.44
[5]D.G.Thompson,J.S.Bates.Range Dependent Phase Gradient Autofocus.IEEE Trans.On GRS,July.1998,vol.5:2634-2636.
[6]RLacomme,J.P.Hardange.Air and Spaceborne Radar System:An Introduction.William Andrew Publishing,2001
[7]李燕平、邢孟道、保铮.一种改进的相位梯度自聚焦算法..西安电子科技大学学报(自然科学版)2007,34(3):386-391.
[8]M.Y.Jin,F.Cheng,M.Chen.Chirp scaling algorithms for SAR processing.IEEE Trans.On GRS,Aug.1993,Vol.3:1169-1172.
[9]G.W.Stimson.机载雷达导论.北京:电子工业出版社,2005
[10]张澄波,综合孔径雷达:原理、系统分析与应用,北京:科学出版社,1989.
[11]刘永坦,雷达成像技术,哈尔滨:哈尔滨工业大学出版社,1999.
[12]保铮、邢孟道、王彤,雷达成像技术,北京:电子工业出版社,2005.
[13]张明友、汪学刚,雷达系统,北京:电子工业出版社,2006.
[14]季节,世界机载雷达手册,北京:航空工业大学出版社,1989.10。
[15]毛士艺 著.脉冲多普勒雷达,北京:国防工业出版社,1990.2。
[16]冉承其、朱炬波、王正明SAR图像分辨率的统计判别准则 现代雷达,2005年4月 第27卷 第4期
[17]徐玉芬,线性调频信号数字脉压的分析及其实现系统.现代雷达,1999,21(1):39-43.
[18]Henri Maitre编、孙洪等译合成孔径雷达图像处理.北京:电子工业出版社,2005.2
[19]S.J.Lim.Two-Dimensional Signal and Image Processing.Englewood Cliffs,NJ:Prentice Hall,1990
[20]Xingsong Hou,Guizhong Liu,and Yiyang Zou,SAR Image Data Compression Using Wavelet Packet Transform and Universal-Trellis Coded Quantization.IEEE Trans.On GRS,Nov.2004 Page(s):2632-2641
[21]W.Hughes,K.Gault,G.J.Princz.A Comparison of the Range-Doppler and Chirp Scaling Algorithm with Refence to RADARSAT,IEEE Trans.On GRS,May.1996,Vol.2:1221-1223.
[22]Aria B.Ramirez,Ivan J.Rivera.SAR Image Processing Algorithms Based on the Ambiguity Function.IEEE 7-10 Aug.2005 Page(s):1430-1433 Vol.2
[23]罗军辉、罗勇江、白义臣等编著.《MATLAB7.0在数字信号处理中的应用》[M],北京:机械工业出版社,2005.5。
[24]陈小蔷,合成孔径雷达自聚焦算法研究,南京:南京航空航天大学硕士论文,2001.4
[25]王越,合成孔径雷达自聚焦算法研究,南京:南京航空航天大学硕士论文,2006.12
[26]陶青年,SAR/ISAR基于回波信号的运动补偿技术研究,南京,南京航空航天大学硕士论文,2005.
[25]G.W.Davidson.The Effect of Pulse Phase Errors on the Chirp Scaling SAR Processing Algorithm.IEEE Trans.On GRS,Mar.1996,Vol.34,No.2.
[26]雷万明、胡学成,基于回波数据的高分辨率机载SAR运动补偿,电子与信息学报,2004,26(12):1908-1914.
[27]李玺、倪晋麟、刘国岁等,基于图像准则的SAR/ISAR相位补偿技术的研究.电子科学学刊,2000,22(2):279-289.
[28]解培中,合成孔径雷达位移和相关法自聚焦的研究.南京电子工程研究中心,1995.
[29]武昕伟、朱兆达,利用对比度最大化实现SAR图像自聚焦.现代雷达,2002,24(3):20-22.
[30]马仑、廖桂生,一种基于最大全变差准则的SAR图像自聚焦算法.西安电子科技大学学报(自然科学版),2006,33(2):169-172.
[31]叶少华、周荫清、夏义华,与图像对比度准则相结合的PGA自聚焦算法.现代雷达,2004,26(6):26-32.
[32]康雪艳,对比度最优与子孔径相关自聚焦算法的比较,测试技术学院,2003,17(1):19-24.
[33]张新、吴一戎、丁赤飚,一种易行的高分辨率机载SAR实时自聚焦算法,电子与信息学报,2006,28(5):923-926.
[34]J.S.Bates.The Phase Gradient Autofocus Algorithm with Range Dependent Stripmap SAR.England:Brigham Young University,1998.
[35]雷万明、胡学成,基于PGA的多普勒调频斜率估计,电子学报,2004,32(6):1017-1019.
[36]李力、Asif Raza、毛士艺,秩—相位误差估计自聚焦方法的改进,电子科学学刊,1999,26(9):771-778.
[37]F.M.Dickey,L.A.Romero.Super-resolution,Degrees of Freedom and Synthetic Aperture Radar.IEE Proc.Radar Sonae Navig,2003,150(6):419-429
[38]陈琦、李景文,相位梯度自聚焦算法的性能分析及改进,北京航空航天大学学报,2004,30(2):131-134.
[39]武昕伟、朱兆达,一种基于最小熵准则的SAR图像自聚焦算法,现代雷达,2003,25(7):437-440.
[40]D.Wenzel,.Mittermayer,A.Moreira.Analysis of Two Different Implementations of the SPECAN Algorithm for Real-time Range Compression of SAR Data.IEEE Trans.On GRS,July.1999,vol.3:1764-1766.
[41]Jesper Schou,Henning Skriver,and Knut Conradsen,ICFAR Edge Detector for Polarimetric SAR Images.IEEE Trans.On GRS,Jan.2003,vol.41,no.1.
[42]Rafael C.Gonzalez、Richard E.Woods著 阮秋琦 阮宇智 等译.数字图像处理(第二版)北京:电子工业出版社2003年3月
[43]卢健、卢昕.高分辨率机载SAR图像与光学图像的融合及应用分析.2005年中国合成孔径雷达会议论文集
[44]裘金飞.合成孔径雷达(SAR)成像技术研究 2005年6月 南京理工大学硕士学位论文
[45]刘月花.高分辨率机载SAR信号处理研究 中国科学院硕士学位论文 2001年6月
[46]J.M.Horrell,M.R.Inggs,Satellite and Airborne.SAR Simulator,IEEE Trans.On GRS,Aug.1993 Page(s):193-198
[47]Woifgang-Martin Boerner,Recent Advances In Polarimetric-Interferometric SAR Throry &Technology And Its Application,IEEE Trans.On GRS,May.2000 Page(s):212-229
[48]赵侠、汪雄良、王正明等.基于最优对比度准则的SAR图像相位梯度自聚焦算法.遥感技术与应用,2005,Dec.20(6),pp:606-610
[2]王正明、朱炬波等著.SAR图像提高分辨率技术.北京:科学出版社,2006.10
[3]张直中.雷达信号的选择与处理.北京:国防工业出版社,1989
[4]J.Li,P.Stoica.An Adaptive Filtering Approach to Spectral Estimation and SAR Imaging.IEEE Signal Processing,June 1996,Vol.44
[5]D.G.Thompson,J.S.Bates.Range Dependent Phase Gradient Autofocus.IEEE Trans.On GRS,July.1998,vol.5:2634-2636.
[6]RLacomme,J.P.Hardange.Air and Spaceborne Radar System:An Introduction.William Andrew Publishing,2001
[7]李燕平、邢孟道、保铮.一种改进的相位梯度自聚焦算法..西安电子科技大学学报(自然科学版)2007,34(3):386-391.
[8]M.Y.Jin,F.Cheng,M.Chen.Chirp scaling algorithms for SAR processing.IEEE Trans.On GRS,Aug.1993,Vol.3:1169-1172.
[9]G.W.Stimson.机载雷达导论.北京:电子工业出版社,2005
[10]张澄波,综合孔径雷达:原理、系统分析与应用,北京:科学出版社,1989.
[11]刘永坦,雷达成像技术,哈尔滨:哈尔滨工业大学出版社,1999.
[12]保铮、邢孟道、王彤,雷达成像技术,北京:电子工业出版社,2005.
[13]张明友、汪学刚,雷达系统,北京:电子工业出版社,2006.
[14]季节,世界机载雷达手册,北京:航空工业大学出版社,1989.10。
[15]毛士艺 著.脉冲多普勒雷达,北京:国防工业出版社,1990.2。
[16]冉承其、朱炬波、王正明SAR图像分辨率的统计判别准则 现代雷达,2005年4月 第27卷 第4期
[17]徐玉芬,线性调频信号数字脉压的分析及其实现系统.现代雷达,1999,21(1):39-43.
[18]Henri Maitre编、孙洪等译合成孔径雷达图像处理.北京:电子工业出版社,2005.2
[19]S.J.Lim.Two-Dimensional Signal and Image Processing.Englewood Cliffs,NJ:Prentice Hall,1990
[20]Xingsong Hou,Guizhong Liu,and Yiyang Zou,SAR Image Data Compression Using Wavelet Packet Transform and Universal-Trellis Coded Quantization.IEEE Trans.On GRS,Nov.2004 Page(s):2632-2641
[21]W.Hughes,K.Gault,G.J.Princz.A Comparison of the Range-Doppler and Chirp Scaling Algorithm with Refence to RADARSAT,IEEE Trans.On GRS,May.1996,Vol.2:1221-1223.
[22]Aria B.Ramirez,Ivan J.Rivera.SAR Image Processing Algorithms Based on the Ambiguity Function.IEEE 7-10 Aug.2005 Page(s):1430-1433 Vol.2
[23]罗军辉、罗勇江、白义臣等编著.《MATLAB7.0在数字信号处理中的应用》[M],北京:机械工业出版社,2005.5。
[24]陈小蔷,合成孔径雷达自聚焦算法研究,南京:南京航空航天大学硕士论文,2001.4
[25]王越,合成孔径雷达自聚焦算法研究,南京:南京航空航天大学硕士论文,2006.12
[26]陶青年,SAR/ISAR基于回波信号的运动补偿技术研究,南京,南京航空航天大学硕士论文,2005.
[25]G.W.Davidson.The Effect of Pulse Phase Errors on the Chirp Scaling SAR Processing Algorithm.IEEE Trans.On GRS,Mar.1996,Vol.34,No.2.
[26]雷万明、胡学成,基于回波数据的高分辨率机载SAR运动补偿,电子与信息学报,2004,26(12):1908-1914.
[27]李玺、倪晋麟、刘国岁等,基于图像准则的SAR/ISAR相位补偿技术的研究.电子科学学刊,2000,22(2):279-289.
[28]解培中,合成孔径雷达位移和相关法自聚焦的研究.南京电子工程研究中心,1995.
[29]武昕伟、朱兆达,利用对比度最大化实现SAR图像自聚焦.现代雷达,2002,24(3):20-22.
[30]马仑、廖桂生,一种基于最大全变差准则的SAR图像自聚焦算法.西安电子科技大学学报(自然科学版),2006,33(2):169-172.
[31]叶少华、周荫清、夏义华,与图像对比度准则相结合的PGA自聚焦算法.现代雷达,2004,26(6):26-32.
[32]康雪艳,对比度最优与子孔径相关自聚焦算法的比较,测试技术学院,2003,17(1):19-24.
[33]张新、吴一戎、丁赤飚,一种易行的高分辨率机载SAR实时自聚焦算法,电子与信息学报,2006,28(5):923-926.
[34]J.S.Bates.The Phase Gradient Autofocus Algorithm with Range Dependent Stripmap SAR.England:Brigham Young University,1998.
[35]雷万明、胡学成,基于PGA的多普勒调频斜率估计,电子学报,2004,32(6):1017-1019.
[36]李力、Asif Raza、毛士艺,秩—相位误差估计自聚焦方法的改进,电子科学学刊,1999,26(9):771-778.
[37]F.M.Dickey,L.A.Romero.Super-resolution,Degrees of Freedom and Synthetic Aperture Radar.IEE Proc.Radar Sonae Navig,2003,150(6):419-429
[38]陈琦、李景文,相位梯度自聚焦算法的性能分析及改进,北京航空航天大学学报,2004,30(2):131-134.
[39]武昕伟、朱兆达,一种基于最小熵准则的SAR图像自聚焦算法,现代雷达,2003,25(7):437-440.
[40]D.Wenzel,.Mittermayer,A.Moreira.Analysis of Two Different Implementations of the SPECAN Algorithm for Real-time Range Compression of SAR Data.IEEE Trans.On GRS,July.1999,vol.3:1764-1766.
[41]Jesper Schou,Henning Skriver,and Knut Conradsen,ICFAR Edge Detector for Polarimetric SAR Images.IEEE Trans.On GRS,Jan.2003,vol.41,no.1.
[42]Rafael C.Gonzalez、Richard E.Woods著 阮秋琦 阮宇智 等译.数字图像处理(第二版)北京:电子工业出版社2003年3月
[43]卢健、卢昕.高分辨率机载SAR图像与光学图像的融合及应用分析.2005年中国合成孔径雷达会议论文集
[44]裘金飞.合成孔径雷达(SAR)成像技术研究 2005年6月 南京理工大学硕士学位论文
[45]刘月花.高分辨率机载SAR信号处理研究 中国科学院硕士学位论文 2001年6月
[46]J.M.Horrell,M.R.Inggs,Satellite and Airborne.SAR Simulator,IEEE Trans.On GRS,Aug.1993 Page(s):193-198
[47]Woifgang-Martin Boerner,Recent Advances In Polarimetric-Interferometric SAR Throry &Technology And Its Application,IEEE Trans.On GRS,May.2000 Page(s):212-229
[48]赵侠、汪雄良、王正明等.基于最优对比度准则的SAR图像相位梯度自聚焦算法.遥感技术与应用,2005,Dec.20(6),pp:606-610